Identification of functional variants is an end-game of analysis of genetically influenced diseases and a starting point to understand roles of genes in behaviors. Public databases are populated with >22 million sequence variants, mostly single nucleotide polymorphisms. However, most rare and uncommon variants are unknown, and functionality of most is unknown. Through in-vivo and in vitro functional genomics studies we have discovered several loci with roles in complex behaviors relevant to alcoholism and addictions. These studies make use of humans, including individuals from founder populations and who experienced particular exposures or were evaluated for intermediate phenotypes, and model organisms, including mice, rats and macaques We discovered or helped define effects of a series of functional polymorphisms altering behavior. In alcoholism, an OPRM1 Asn40Asp missense variant (Bergen et al, 1997) was shown by others to be functional and linked to naltrexone treatment response (Anton et al, 2008). A common, functional LA->LG SNP in the serotonin transporter HTTLPR locus (Hu et al) enhanced linkage to behaviors and intermediate phenotypes. These include SSRI response of depressed patients (Hu et al, 2007), neuroimaging responses to emotion, and gene x stress interactions leading to suicidality (Roy et al, 2007). Common HTR2C Ser23Cys (Lappalainen et al) and HTR2A Asn452His alleles (Ozaki et al) were detected and shown to be functional, and linked by others to behavior, including clozapine response. In first-episode schizophrenics, we found that a functional DRD2 promoter polymorphism influences antipsychotic response (Lencz et al, 2006). We traced linkages of functional loci and haplotypes of NPY (Zhu et al, 2008), GCH1 (Tegeder et al, 2006) and DISC1 (a series of papers beginning with Hodgkinson et al, 2004) to various behaviors including emotionality, schizophrenia and clinical pain outcome. By deep sequencing we found an HTR2B Stop codon that is relatively common in Finns, absent in other populations, and associated with but not determinant for impulsive behavior. This stop codon also cosegregated with impulsive behavior in families, and the mouse gene knockout was more impulsive and higher in novelty seeking. As an example of the type of molecular studies performed with polymorphisms listed in this paragraph, the HTR2B stop codon led to variable nonsense mediated decay of the HTR2B RNA and blocked expression of the receptor (Bevilacqua et al, Nature, 2010). A low expression Neuropeptide Y (NPY) haplotype increased anxiety and emotionality but had stronger effects on molecules (NPY RNA and protein) and intermediate phenotypes (brain imaging responses to emotion and pain/stress) than complex behavior(Zhou et al, Nature, 2008). Intermediate phenotypes augment analysis of behavior. The imaging genetics paradigm we helped initiate (Heinz et al, 2000; Hariri et al, 2002; Egan et al, 2001; 2003; Zubieta et al, 2003) led to groundbreaking findings. With NIDA collaborators, we helped discover a mechanism of action of CHRNA5 Asn398, the major functional locus from nicotine GWAS. The risk allele weakens brain connectivities including a Dorsal Anterior Cingulate/Ventral Striatal circuit that predicts nicotine craving (Hong et al, 2010). Clinical subphenotyping enabled linkage of HTR1B to antisocial alcoholism (Lappalainen et al), serotonin transporter (SLCA4) to anxiety (Mazzanti et al; Hariri et al), BDNF Val66Met to episodic memory (Egan et al), COMT Val158Met to anxiety (Enoch et al), executive cognition (Egan et al; Lipsky et al; Malhotra et al), and pain threshold (Zubieta et al; Diatchenko et al), and GTP cyclohydrolase to chronic pain and experimental pain response (Tegeder et al). Frontal cognitive deficit is a risk factor in schizophrenia, alcoholism and other diseases. Dopamine generally enhances prefrontal cortical efficiency. Met158, a common COMT variant, leads to four-fold reduction in COMT activity. It is thus a candidate allele for cognitive function via effect on frontal dopamine. We found an allele-dosage relationship of Met158 to frontal cognitive function and diminished cortical efficiency (Egan et al). The effect on cognition is observed in populations differing in baseline cognitive function: schizophrenia, moderate-severe head injury (Lipsky et al), & controls (Malhotra et al). We proposed that Val158 has a counter-advantage: stress resiliency. In two populations Met158 predicted anxiety in women and decreased frontal EEG coherence (Enoch et al), and Met158 was associated with lower resiliency to pain/stress (Zubieta et al; Diatchenko et al). Met158 predicted inability to activate endomorphin release after pain/stress (Zubieta et al). Overall, effect sizes of genes in intermediate phenotypes is larger (Goldman and Ducci, 2007) and as illustrated by effects of NPY (Zhou et al, 2008), COMT (Zubieta et al, 2003) and CHRNA5 (Hong et al, 2010) on intermediate phenotypes. A functional polymorphism of Fatty Acid Amide Hydrolase (FAAH) which synthesizes the endogenous ligand for cannabinoid receptors, has specifically predicts placebo response to pain, as shown by our imaging genetic study (Pecina et al, 2014). Linkage, GWAS and RNA-seq are genomic, hypothesis-free approaches. The Chr 4 GABAA subunit cluster we implicated in alcoholism by family linkage (Long et al), an effect that appears anxiety-modulated (Enoch et al). Another GABAA gene cluster implicated in alcoholism and alcohol response is located on Chr 5 (Radel et al). GABRA6 has a missense variant associated with alcohol dependence and response to alcohol and diazepam (Iwata et al, Schuckit et al). A family linkage scan yielded genome-wide significant linkage of CRH-BP (a stress-related protein) to an alcoholism-associated EEG trait and this was followed by association of CRH-BP to EEG in two populations. (Enoch et al). Our GWAS of EEG (Hodgkinson et al, PNAS, 2010) in a relatively small sample of Plains Indians detected three independent genome-wide significant loci(Hodgkinson et al, PNAS 2010). Founder populations, extreme probands, and exposed populations enhance power to detect gene effects. Our Finnish dataset is derived from a founder population and ascertained from criminal alcoholic probands & thus enriched for Type II early onset alcoholism. SW and Plains Indian samples represent founder populations. An African American substance dependence sample with high adversity exposure revealed GxE of childhood adversity and HTTLPR in suicidality (Roy et al, 2007). Stress and poverty, but not African ancestry, predicted high risk of addictions (Ducci et al, 2009). An MAOA VNTR previously linked to dyscontrol via stress interaction was linked to outcomes of alcohol dependence and ASPD in American Indian women, of whom half had been sexually abused as children (Ducci et al, 2008). A strong interaction between the MAOA VNTR and testosterone was observed in the Finnish criminal sample (Sjoberg et al). As mentioned, deep sequencing in Finns detected a population-specific HTR2B Stop codon significant for impulsivity (Bevilacqua et al, Nature, 2010). We exome-sequenced an artificially selected animal model for alcoholism, the P/NP rat, and discovered a stop codon in the metabotropic glutamate receptor 2 (Grm2) gene that leads to altered glutamate function and increased alcohol preference in the alcohol-preferring rat(Zhou et al, 2013). The Grm2 Stop codon is relatively common (0.08) in parental Wistar rats and leads to widespread, uncompensated changes in glutamate function. However, it has been genetically fixed in the artificially selected strains illustrating the power of using such strains to discover alleles that alter behavior.